[1]印 峰,王耀南,杨易旻,等.结合变尺度法的改进类电磁机制算法[J].智能系统学报,2010,5(3):254-259.
YIN Feng,WANG Yao-nan,YANG Yi-min,et al.An improved electromagnetismlike mechanism algorithmcombined with the DFP method[J].CAAI Transactions on Intelligent Systems,2010,5(3):254-259.
点击复制
《智能系统学报》[ISSN 1673-4785/CN 23-1538/TP] 卷:
5
期数:
2010年第3期
页码:
254-259
栏目:
学术论文—人工智能基础
出版日期:
2010-06-25
- Title:
-
An improved electromagnetismlike mechanism algorithmcombined with the DFP method
- 文章编号:
-
1673-4785(2010)03-0254-06
- 作者:
-
印 峰,王耀南,杨易旻,曹文明
-
湖南大学 电气与信息工程学院,湖南 长沙 410082
- Author(s):
-
YIN Feng, WANG Yao-nan, YANG Yi-min, CAO Wen-ming
-
College of Electrical and Information Engineering, Hunan University, Changsha 410082, China
-
- 关键词:
-
连续函数优化; 类电磁算法; 变尺度法; 二次优化
- Keywords:
-
continuous function optimization; electromagnetismlike algorithm; DavidonFletcherPowell method; quadratic optimization
- 分类号:
-
TP301.6
- 文献标志码:
-
A
- 摘要:
-
标准类电磁机制算法处理连续函数优化问题时存在最优参数选取和收敛速度问题.数值实验研究表明类电磁算法不具备初值敏感性,并在搜索后期算法收敛速度缓慢甚至可能出现停滞.数值实验分析指出粒子之间达成动态力平衡状态是造成算法停滞的可能原因之一,提出一种解决策略是摒弃EM算法后期搜索过程,结合变尺度法对EM算法前期搜索到的近似最优值进行二次优化.该混合计算方法将二者的优势相结合,实验结果表明新方法在保证计算实时性的同时,取得了较高的计算精度.最后,对EM算法本身构造提出一些改进意见,并初步建立用于连续函数优化的EM算法计算框架,为后续更深入的研究EM算法提供参考.
- Abstract:
-
When optimizing a continuous function using a standard electromagnetismlike mechanism (EM), there are known problems, including selection of optimal parameters and convergence speed. Numerical simulations indicate that EM is not sensitive to initial values, but algorithm convergence is slow and may even stagnate in the latter part of a search. One of the possible causes for stagnation is the equilibrium of dynamic forces between particles. In order to improve the performance of the EM, instead of the latter search process, a quadratic optimization method was proposed. When combined with the DavidonFletcherPowell (DFP) method, it optimized the approximate optimal results obtained by a presearch with the EM algorithm. This hybrid method fully exploits the strengths of the EM method and the DFP method. Experimental results showed it to be more efficient and precise. Finally, some improvements were made to the construction of the EM and a framework for the EM method was established that allows continuous function optimization. These results provide a reference for more indepth study of the EM algorithm.
备注/Memo
收稿日期:2009-12-15.
基金项目:国家科技支撑计划资助项目(2008BAF36B01);国家“863”计划资助项目(2008AA04Z214)
通信作者:印 峰.
E-mail:yinfeng83@126.com.
作者简介:
印 峰,男,1983年生,博士研究生,主要研究方向为机器人控制及人工智能计算.
王耀南,男,1957年生,教授、博士生导师,湖南大学电气与信息工程学院院长,国家高效磨削工程技术研究中心副主任,教育部输变电新技术工程研究中心主任,国际IEEE高级会员,国际自动控制联IFAC会员,中国人工智能学会、自动化学会、电机工程学会理事.主要研究方向为智能机器人、智能信息处理和智能控制.目前在研国家级科研项目共5项,获国家科学技术进步二等奖3项,湖南科学技术进步一等奖等其他奖励多项.发表学术论文300余篇,出版专著和教材4部.
?杨易旻,男,1982生,博士研究生,主要研究方向为机器人智能控制、智能信息处理.
更新日期/Last Update:
2010-08-27